Journal of Environmental Sciences 20(2008) 1323–1327

Long-term monitoring and source apportionment of PM2.5/PM10

in Beijing, China

WANG Hailin1, ZHUANG Yahui1, WANG Ying2, SUN Yele2,YUAN Hui2, ZHUANG Guoshun2, HAO Zhengping1,∗

1. Research Center for Eco-Environmental Sciences, The Chinese Academy of Sciences, Beijing 100085, China. E-mail: wanghailin@rcees.ac.cn2. Center for Atmospheric Environmental Study, Department of Chemistry, Beijing Normal University, Beijing 100875, China

Received 24 December 2007; revised 1 February 2008; accepted 15 April 2008

AbstractDuring 2001–2006, PM2.5 (particle matter with aerodynamic diameter less than 2.5 microns) and PM10 (particle matter with

aerodynamic diameter less than 10 microns) were collected at the Beijng Normal University (BNU) site, China, and in 2006, at a

background site in Duolun (DL). The long-term monitoring data of elements, ions, and black carbon showed that the major constituents

of PM2.5 were black carbon (BC) crustal elements, nitrates, ammonium salts, and sulfates. These five major components accounted for

20%–80% of the total PM2.5. During this period, levels of Pb and S in PM remained rather high, as compared with the levels in other

large cities in the world. Source apportionment results suggest that there were 6 common sources for PM2.5 and PM10, i.e., soil dust,

vehicular emission, coal combustion, secondary aerosol, industrial emission, and biomass burning. Coal combustion was the largest

contributor of PM2.5 with a percentage of 16.6%, whereas soil dust played the most important role in PM10 with a percentage of 27%.

In contrast, only three common types of sources could be resolved at the background DL site, namely, soil dust, biomass combustion,

and secondary aerosol from combustion sources.

Key words: PM2.5; PM10; monitoring; source apportionment

Introduction

There has been growing concern on the high levels of

airborne particulate matter in Beijing in the recent years,

and inhalable particulate has long been implicated to have

adverse health impacts such as respiratory diseases and

increased mortality (Dockery and Pope III, 1994; Donald-

son et al., 1998). Atmospheric particles through long-range

transportation from China mainland to the Pacific can

affect the global radiation energy balance (Liu and Peter,

2002) as well as influence the primary productivity of the

Ocean and the global carbon budget (Zhuang et al., 2001)According to the data of Environmental Situation Bulletin

(BJEPB, 2008), all the annual average concentrations of

inhalable paritcles PM10 (particle matter with aerodynamic

diameter less than 10 microns) in Beijing over the period

of 1999–2006 exceed the national second-class air quality

standard of 100 mg/m3 (MEP, 1996). A number of articles

have been published in the recent years dedicated to the

characterization of particulate matter in Beijing (He etal., 2001; Sun et al., 2004). However, their monitoring

activities usually lasted for only one or two years. Limited

samples and short monitoring time made deliberate source

identification and trend estimation difficult. Systematic

and long term records of particulate matter mass and

* Corresponding author. E-mail: zpinghao@rcees.ac.cn.

composition for a developing country like China will help

to enhance and create scientific knowledge necessary to

address the worsening air quality. In particular, it will help

to understand the nature of the particulate pollution in

a city in relation to local sources, long-range transport,

and atmospheric transformation processes that is essential

for the formulation of effective air quality management

strategies.

In this article, we report the results of PM charac-

terization over a period of six years, overlapping the

implementation period of major air pollution control mea-

sures in Beijing. Although several studies on PM source

apportionment have been conducted in Beijing (Song et al.,2007; Zhang et al., 2007), the lack of an ideal background

site as well as the lack of long-term monitoring dataset

were the common problems. Our source apportionment

was based not only on our six-year long-term dataset at

an urban sampling site at the Beijng Normal University

(BNU), but also on a comparison with the data from a

remote background site in Duolun (DL), about 250 km

north in the upwind direction of Beijing, where the sand

dunes are the nearest to Beijing and threathen its air

quality. Such a comparison can throw some light on the

differentiation of local soil dust and the dust transported

from remote sources like DL.

1324 WANG Hailin et al. Vol. 20

1 Materials and methods

1.1 Sampling and analysis

The PM2.5/PM10 samples were collected during the sum-

mertime and wintertime (about one month, respectively)

from 2001 to 2006 on the roof of the building of Science

and Technology in Beijing Normal University (BNU),

which is about 40 m high. The BNU campus is located

near a main arterial road of Beijing – the North Third Ring

Road. In 2006, PM2.5/PM10 samples were also collected

at a remote background site in Duolun, which is located

about 250 km north from the center of Beijing and is in its

upwind direction during the dry season.

Previously calibrated medium volume samplers (Beijng

Geological Instrument-Dicked Co., Ltd., China) were em-

polyed for PM2.5 and PM10 collection at a constant flow

of 77.59 L/min. The sampling time lasted for about 12 h

for each sample. Mixed fibre filters (Whatman 41, UK)

were used in this campaign for mass, BC, and inorganic

chemical species determinations. After sampling, these

filters were immediately placed in polyethylene plastic

bags and stored in refrigerator, and were then weighed

under constant temperature (20 ± 5°C) and humidity (40%

± 2%), using an analytical balance (Sartorius 2004MP,

Germany) with a reading precision of 10 μg.

An inductively coupled plasma spectroscopy and atomic

emission spectroscopy (ICP-AES) (Ulttma, Jobin Yvon

Co., Ltd., France) were used for the determinations of 23

elements, including Al, As, Ca, Ce, Cd, Cr, Cu, Co, Eu,

Fe, Mg, Mn, Na, Ni, Pb, S, Sc, Sr, Se, Sb, Ti, V, and

Zn (Zhuang et al., 2001). Ionic Chromatograph (Model

600, Dionex, USA) was empoyed for the measurements of

10 anions (Cl−, F−, SO42−, NO3

−, C2O42−, PO4

3−, NO2−,

HCOO−, MSA, CH3COO−) and 5 cations (Ca2+, Mg2+,

Na+, K+, NH4+) according to the procedure developed by

Yuan et al. (2003). A smoke stain reflectometer (Model

43D, Diffusion Systems Ltd., UK) was employed for black

carbon (BC) analysis.

Data quality assurance, including inter-laboratory com-

parison of Standard Reference Material (SRM) measure-

ments and sampler performance, has been taken into

account.

1.2 PMF modeling description

The technique of positive matrix factorization (PMF),

developed by Paatero and Tapper (1994), was adopted for

source apportionment. Owing to the non-negative factor

results and appropriate error estimations, it has been wide-

ly used in recent years (Anttila et al., 1995; Lee et al.,1999; Hedberg et al., 2005; Rizzo and Scheff, 2007). The

general receptor model of PMF can be written as Eq.(1).

X = G × F + E (1)

where, X is the n × m matrix of ambient element concen-

trations, G is the n × p matrix of source contributions,

F is the p × m matrix of source profiles, and E is the

matrix of residuals. The matrices, G, F, and E are estimated

from the known matrix X in PMF. To find the right source

numbers, it is necessary to test different sources with

different settings of parameters and find an optimal one

giving the most physically reasonable results.

Since the PM level in Beijing is generally quite high,

sampling periods were constrained to less than 24 h and

were not consistent. Thus, all raw data (PM2.5/PM10) were

normalized to a 24-h basis before receptor modeling.

In the input data set, the data for species with below

detection limit (BDL) values were replaced with values

corresponding to one-half of the appropriate analytical

detection limit, and the missing species data were replaced

with values corresponding to four times of the arithmetic

mean of relative species.

After excluding species that were frequently present

at concentrations below BDL values, 24 species out of

36 species were chosen for the BNU PM2.5/PM10 factor

analysis: BC, NO3−, SO4

2−, Cl−, C2O42−, NH4

+, K+, Na,

Mg, Al, Ca, Ti, V, Cr, Mn, Fe, Ni, Cu, Zn, As, Cd, Co, S,

and Pb. The total number of samples was 216 and 223 for

the BNU PM2.5 and PM10 analysis, respectively, and the

DL PM2.5/PM10 samples were pretreated in the same way.

2 Results and discussion

2.1 Trend of seasonal variation of PM2.5/PM10 levels inBeijing in 2001–2006

As seen in Fig.1, the PM2.5 levels in Beijing in these

years remained rather high, and are considerably higher

than the average daily value of 65 μg/m3 recommended by

United States Environmental Protection Agency (USEPA).

However, it is interesting to note that the PM2.5 concentra-

tion in wintertime has decreased somewhat from 2001 to

2003 and has retained a steady level since 2004. In view

of the soaring increase in urban population and vehicular

population in these years, the steady PM2.5 level in recent

years implies that the fuel shift from coal to natural gas for

space heating in urban areas may compensate the incre-

ment in vehicular emissions. On the contrary, longer and

stronger solar irradiation during summer time can favor the

photochemical formation of secondary aerosol particles,

which are the major constituents of PM2.5 (Sun, 2005) that

lead to the higher PM2.5 levels during summertime.

Fig. 1 PM2.5 seasonal variation at site of Beijng Normal University

(BNU), China.

No. 11 Long-term monitoring and source apportionment of PM2.5/PM10 in Beijing, China 1325

PM2.5 is of more concern in this study, since there is

no routine monitoring of PM2.5 in China and it has larger

health impacts. Generally speaking, the levels of Pb in

PM2.5 decreased abruptly since 2001, indicating that the

efforts of phasing out leaded gasoline did work (Fig.2).

Nevertheless, the daily levels of Pb fluctuated significantly

from 7.1 to 2,370 ng/m3 with a mean of 180 ng/m3

(Sdandard Division 233). Although they are lower than the

figures in the previous reports (Sun et al., 2006), they are

considerably higher than the values in other cities (Chow

et al., 1996; Gao et al., 2002; Huang et al., 1996). In view

that leaded gasoline has been banned since 1997 in Beijing,

there may be some unknown minor Pb sources, such as

illegal leaded gasoline used in vehicles from remote cities,

industrial activities, coal combustion, and re-entrainment

of lead-contaminated soil particles. Further investigation

is needed for clear source identification.

Regarding the pollution element As, a descending trend

was observed (Fig.2), which may be explained by the

reduction of coal consumption in the urban areas, as coal

combustion is the main source of As in Beijing. However,

coal is still used in the suburban areas and in nearby cities,

therefore, As can still be detected.

There seems to be a good correlationship between the

total sulfur content and the sulfate content in PM2.5 (R2

= 0.68). It implies that the major source of sulfur is coal

combustion. High concentrations of S frequently appear in

wintertime (Fig.2).

The ratio of NO3−/SO4

2− has been used in China as

an indicator of the relative importance of mobile source

vs. stationary coal combustion source (Wang et al., 2005).Arimoto et al. (1996) ascribed high NO3

−/SO42− ratios to

the predominance of mobile source over stationary source

of pollutants. Here, the average ratios of NO3−/SO4

2− in

PM2.5 varied from 0.34 to 1.11 with an average value of

0.63 (data not shown). There is a slow increasing trend of

the NO3−/SO4

2− ratio, suggesting that the contribution of

mobile emissions in Beijing is increasing. However, the

ratio in general retained a value less than 1. This implies

that coal combustion still remains to be a primary emission

source.

Fig. 2 Temporal variation of important pollutants in PM2.5 at the BNU

site during 2001–2006. (a) 2001 summer; (b) 2001 winter; (c): 2002

summer; (d) 2002 winter; (e) 2003 winter; (f) 2004 winter; (g) 2005

summer; (h) 2005 winter; (i) 2006 summer; (j) 2006 winter.

2.2 Source apportionment of Beijing PM2.5/PM10 byPMF

We performed testing runs with different number of

factors to find an optimal one with the most physically

meaningful results. The results given by 8 and 6 factors

for PM2.5 and PM10, respectively, are the most reasonable

and easy for interpretation. An important output is the

explained variation (EV), which is dimensionless and

indicates how important each factor is in explaining the

observed matrix. The value of EV ranges from 0.0 to 1.0,

from no explanation to complete explanation. In the results

of the BNU data, about 86.3% of the observed PM2.5 data

can be explained by the eight factors, and 88.6% of the

observed PM10 data can be explained by the six factors.

The following are the common factors for both PM2.5 and

PM10:

Source 1 represents the local soil-related sources with

typical crust elements, i.e., Al, Fe, Mn, Mg, and Ca,

including air-borne road dust and fugitive dust. The high

EV value of Ca may be associated with the intensive

construction activities in Beijing in recent years. This is

an important source of Beijing aerosol.

Source 2 is dominated by SO42−, NH4

+, and NO3−,

representing the sources of secondary ionic particles. The

secondary sulfates are formed by photochemical reactions,

especially in summer with strong solar radiation and high

temperature (Seinfield and Pandis, 1998). It was also

reported that high humidity will promote the formation of

secondary sulfates (Yao et al., 2003). Nitrates are mainly

formed by the oxidation of NOx. Low temperature will

favor the formation of ammonium nitrate aerosol (Li et al.,2004). This can explain the high contribution of nitrates in

wintertime. For PM2.5, secondary sulfates and secondary

nitrates were resolved as two independent sources, whereas

for PM10, they were emerged as one secondary aerosol

source.

Source 3 is rich in Zn, Pb, Cu, and BC. Zn possibly

comes from the fuel burning or tire abrasion (Pacyna,

1986), exist in the mobile exhaust (Huang et al., 1994). Zn

has been used as the fingerprint of mobile vehicle emission

sources in recent PMF analysis (Li et al., 2004). Cu may

come from the braking system and the pump system (Lee

et al., 1999). Pb may have been a minor additive for

low-grade fuel produced by out-of-dated oil refineries in

China, as leaded-gasoline has been phased out since 1999

in China. In source 3, we also found a medium contribution

of BC. Hence, we ascribe source 3 to the mobile source.

Source 4 is rich in K, representing the source of

biomass-burning emissions. It includes most likely house-

hold combustion of agricultural residues and firewood,

plus little incineration. As we explained earlier, the EC/OC

(elemental carbon/organic carbon) data were rare, and

were not taken simultaneously. For this reason, we did not

include the EC/OC data.

Source 5 is a minor industrial emission source, charac-

terized by As, V, Cd, Mn, and Fe, probably from industrial

smelting emissions. Lee et al. (1999) also reported a simi-

lar factor for the Hong Kong’s PM source apportionment.

1326 WANG Hailin et al. Vol. 20

Source 6 is dominated by BC, S, and Cl. BC and S

usually come from coal combustion. It was also found

that Cl was associated with coal burning during wintertime

(Yao et al., 2002). Therefore, source 6 can be ascribed to

coal combustion.

Besides the above-mentioned six common sources, there

is an unknown source for PM2.5. In this source, Na has the

highest EV value. A similar source having high EV values

of Na and Si has been ascribed to non-local dust (Xie

and Liu, 2007; Zhao and Hopke, 2004, 2006). Since we

used hydrofluoric acid for sample pre-treatment, we did not

have Si data. However, in this factor, we found some other

elements like Cl and S with high EV values correlated

with sodium. The non-local dust in Beijing PM is mainly

transported from Inner Mongolia (Xie and Liu, 2007),

where there are several dried salinas and the main sodium-

containing constituents are Na2SO4·10H2O and NaCl (Xu,

1993; Li et al., 1990). Therefore, particles transported from

these remote sources are characterized by their contents

of sodium sulfate and sodium chloride. Consequently, it

seems reasonable to ascribe this factor as the non-local

dust. In PM10, this source possesses only a small weight,

and cannot be resolved by PMF.

The PM2.5/PM10 samples at the DL site were also used

for source apportionment by PMF, and the results show

fewer sources at this background site, as expected. There

are four factors for DL PM2.5 with 82% explanation: i.e.,

local dust enriched in Mg, Al, Ca, Ti, and Fe, non-local

dust characterized by high Na EV value, biomass burning

characterized by K+, and secondary nitrate represented by

NH4+ and NO3

−. For DL PM10, similar 4 factors pro-

vide the most physically reasonable explanation: namely,

local dust characterized by Ca, Ti, Mn, Fe, Mg, and Al,

non-local dust with high Na EV value, biomass burning

represented by K+, and secondary ions represented by

NH4+, NO3

−, SO42−. In fact, in the fourth source of DL

PM2.5, the secondary nitrate source also contained SO42−.

Since its EV values are considerably lower than those

of nitrates, we ascribed this source to secondary nitrate

instead of secondary ions.

For the quantitative estimation of mass contributions of

the resolved factors, the observed PM mass concentrations

were regressed against the factor scores using MLR (multi-

linear regression), and the average source contribution of

each factor to the total mass can be derived from this

regression process. The annual arithmetic means of the

total mass contribution apportionment for different types

of sources at these two sites are shown in Fig.3. Dust,

combustion activities, and secondary aerosol are the three

major sources of PM2.5/PM10 in Beijing. Among these

sources, coal combustion turned out to be the largest

contributor of PM2.5 with a percentage of 16.7%, whereas

local dust is the primary source of PM10 with a percentage

of 23%. Regarding the PM samples at the DL site, local

dust is the primary source for PM2.5 and PM10 with mass

contributions of 36.2% and 61.7%, respectively.

Fig. 3 Source apportionment for PM2.5/PM10 at the BNU and Duolun

(DL) sites. BB: biomass burning; CC: coal combustion; IE: industrial

emissions; LD: local dust; Non-LD: non-local dust; SI: secondary ions;

SN: secondary nitrate; SS: secondary sulfate; VE: vehicle emissions. Oth-

ers: sources that could not be explained by positive matrix factorization

(PMF).

3 Conclusions

The PM2.5/PM10 samples were collected in summer

and winter during 2001–2006 at the BNU sampling site.

Totally 23 elements, 15 ions, and BC were determined.

The trends of the temporal variations of some important

pollutants in PM2.5 are discussed. Source apportionment

by PMF provides 6 common sources for Beijing PM2.5

and PM10, namely, local dust, secondary aerosol, mobile

source, biomass burning, coal combustion, and minor

industrial emission. Among these sources, soil is the major

source of PM10, while particles of secondary origins are

the main source of PM2.5. Regarding the background site

in Duolun, crustal particles are the dominant constitutent

of both fine and coarse fractions. Biomass combustion is

the most important anthropogenic activity.

Acknowledgments

This work was supported by the National Science Fund

for Distinguished Young Scholars (No. 20725723), the

Swedish International Development Cooperation Agency

(SIDA) and the coordination efforts of Asian Institute of

Technology under the AIRPET program. We are indebted

to Prof. Paatero for giving a license for running the PMF

model.

References

Anttila P, Paatero P, Tapper U, Tapper U, Jarvinen O, 1995.

Source identification of bulk wet deposition in Finland by

positive matrix factorization. Atmos Environ, 29(14): 1705–1718.

Arimoto R, Duce R A, Savoie D L, Prospero J M, Talbot R,

No. 11 Long-term monitoring and source apportionment of PM2.5/PM10 in Beijing, China 1327

Cullen J D et al., 1996. Relationships among aerosol con-

stituents from Asia and the North Pacific durin Pem-West

A. J Geophy Res, 101: 2011–2023.BJEPB (Beijing Municipal Environmental Protection

Bureau), 2008. Environmental Situation Bulletin.

http://www.bjepb.gov.cn/bjhb//tabid/375/Default.aspx.

Dockery D W, Pope III C A, 1994. Acute respiratory effects of

particulate air pollution. Ann Rev Public Health, 15: 107–

132.

Donaldson K, Li X Y, MacNee W, 1998. Ultrafine (nanometer)

particle-mediated lung injury. J Aerosol Sci, 29: 533–560.Chow J C, Watson J G, Lu Z Q, Lowenthal D H, Frazier C A,

Solomon P A, Thuillier R H, Magliano K, 1996. Descriptive

analysis of PM2.5 and PM10 at regionally representative lo-

cations during SJVAQS/AUSPEX. Atmos Environ, 30(12):2079–2112.

Gao Y, Nelson E D, Field M P, Ding Q, Li H, Sherrell R M etal., 2002. Characterization of atmospheric trace elements

on PM2.5 particulate matter over the New York-New Jersey

Harbor Estuary. Atmos Environ, 36(6): 1077–1086.He K B, Yang F M, Ma Y L, Zhang Q, Yao X L, Chan C C, Cadle

S, Chan T, Mulawa P, 2001. The characteristics of PM2.5 in

Beijing, China. Atmos Environ, 35(29): 4959–4970.Hedberg M, Gidhagen L, Johansson C, 2005. Source contribu-

tions to PM10 and arsenic concentrations in Central Chile

using positive matrix factorization. Atmos Environ, 39(3):

549–561.

Huang S, Arimoto R, Rahn K A, 1996. Changes in atmospheric

lead and other pollution elements at Bermuda. J GeophyRes, 101: 21033–21040.

Huang X D, Olmez I, Aras N K, Gordon G E, 1994. Emissions

of trace-elements from motor vehicles-potential marker

elements and source composition profile. Atmos Environ,28(8): 1385–1391.

Lee E, Chan C K, Paatero P, 1999. Application of positive

matrix factorization in source apportionment of particulate

pollutants in Hong Kong. Atmos Environ, 33(19): 3201–

3212.

Li R Q, Zheng L M, Zhu G R, 1990. Innermongolia Lakes

and the Environmental Changes. Beijing: Beijing Normal

University Press.

Li Z, Hopke P K, Husain L, Qureshi S, Dutkiewicz V A, Schwab

J J, Drewnick F, Demerjian K L, 2004. Sources of fine

particle composition in New York city. Atmos Environ,38(38): 6521–6529.

Liu Y, Peter H D, 2002. Anthropogenic aerosols: Indirect warm-

ing effect from dispersion forcing. Nature, 419: 580–581.MEP (Ministry of Environmetal Protection), 1996. Ambient air

quality standard GB 3095-1996. http://www.envir.gov.cn

/law/airql.htm.

Pacyna J M, 1986. Toxic Metals in the Atmosphere. New York:

Willey.

Paatero P, Tapper U, 1994. Positive matrix factorization: a non-

negative factor model with optimal utilization of error

estimates of data values. Environmetrics, 5: 111–126.Rizzo M J, Scheff P A, 2007. Fine particulate source apportion-

ment using data from the USEPA speciation trends network

in Chicago, Illinois: Comparison of two source apportion-

ment models. Atmos Environ, 41(29): 6276–6288.Seinfeld J H, Pandis S N, 1998. Atmospheric chemistry and

physics from air pollution to climate change. New York:

Wiley. 649–655.

Song Y, Tang X Y, Xie S D, Zhang Y H, Wei Y J, Zhang M S

et al., 2007. Source apportionment of PM2.5 in Beijing in

2004. J Hazard Mater, 146(1-2): 124–130.Sun Y L, 2005. Physicochemical characterizations of airborne

particulate matter in the typical regions of China and its

impact on regional environment. Doctoral thesis, Beijing

Normal University, China.

Sun Y L, Zhuang G S, Wang Y, Han L H, Guo J H, Dan M

et al., 2004. The air-borne particulate pollution in Beijing-

concentration, composition, distribution and sources. AtmosEnviron, 38(35): 5991–6004.

Sun Y L, Zhuang G S, Zhang W J, Wang Y, Zhuang Y H, 2006.

Characteristics and sources of lead pollution after phasing

out leaded gasoline in Beijing. Atmos Environ, 40(16):

2973–2985.

Wang Y, Zhuang G S, Tang A H, Yuan H, Sun Y L, Cheng

S, Zheng A H, 2005. The ion chemistry and the source

of PM2.5 aerosol in Beijing. Atmos Environ, 39(21): 3771–3784.

Xie S D, Liu Z, 2007. Source apportionment of inhalable particles

in Beijing. In: Proceeding of Research on the Formation and

Control Technology of Particles from Combustion Sources.

179–195.

Xu X, 1993. Research on the Mineral Compositions of Chinese

Salt Lakes. Beijing: Science Press.

Yao X H, Chan C K, Fang M, Cadle S, Chan T, Mulawa P etal., 2002. The water-soluble ionic composition of PM2.5 in

Shanghai and Beijing, China. Atmos Environ, 36(26): 4223–4234.

Yao X, Lau A P S, Fang M, Chan C K, Hu M, 2003. Size

distribution and formation of ionic species in atmospheric

particulate pollutants in Beijing, China. Atmos Environ,37(21): 2991–3000.

Yuan H, Wang Y, Zhuang G S, 2003. The simultaneous deter-

mination of organic acid, MSA with the inorganic anions

in aerosol and rain-water by ion chromatography. J InstruAnal, 6: 11–14.

Zhang W, Guo J H, Sun Y L, Yuan H, Zhuang G S, Zhuang Y H,

Hao Z P, 2007. Sources apportionment for urban PM2.5 and

PM10 in Beijing area. Chin Sci Bullet, 52(5): 608–615.Zhao W X, Hopke P K, 2004. Souce apportionment for ambient

particles in the San Gorgonio wilderness. Atmos Environ,38(35): 5901–5910.

Zhao W X, Hopke P K, 2006. Source identification for fine

aerosols in Mammoth Cave National Park. Atmos Res,80(4): 309–322.

Zhuang G S, Guo J H, Yuan H, Zhao C Y, 2001. The com-

positions, sources, and size distribution of the dust storm

from China in spring of 2000 and its impact on the global

environment. Chin Sci Bullet, 46(11): 895–901.